Skip to content

docs(repl): factual header, no fixed numbers shown as measurements; ledger benchmark - #11

Merged
quantum-osiris merged 1 commit into
mainfrom
docs/repl-header-and-ledger-bench
Nov 10, 2026
Merged

quantum-osiris merged 1 commit into
mainfrom
docs/repl-header-and-ledger-bench

Conversation

@quantum-osiris

Copy link
Copy Markdown
Owner

Why

The repo is public and is cited in a proposal, so a reviewer may open the console's source or run its status report. Before this change, the module header listed refuted or unsupported items as features: "11D CRSM", θ_lock = 51.843° torsion mechanics, CCCE "consciousness scoring" and a "Merkle cryptographic proof". The introspection report also printed two fixed numbers as if they were measurements.

Changes

osiris_cli/osiris_repl.py

  • The header now says what the module actually wires together. The historical constants are kept, and each one is marked refuted or unsupported, with a pointer to osiris_cli/claims.py.
  • CUSUM Drift Anomaly Score = 0.012 and Strategy Diversity Index = 0.942 were hard-coded numbers. Both lines now say "not measured in this check".
  • Dropped "(Phase-Lock θ=51.843°)" from the coherence line.
  • The argparse description is now plain.

benchmarks/bench_ledger_append.py (new), with its output results_ledger_append.json

  • Measures append latency, verify time and five tamper cases, all on a temporary journal; the tracked journal is never touched.
  • 2,000 appends: median 0.060 ms, p95 0.098 ms, p99 0.164 ms. Verifying all 2,000 entries: 9.3 ms.
  • Detected: editing a field, replacing a stored hash, deleting a middle entry, swapping two neighbouring entries.
  • Not detected: deleting entries from the end. A hash chain alone can't see a removed tail; catching it needs the latest hash and count recorded somewhere an attacker can't rewrite, such as signed on an operator token. The benchmark reports this as it is.

Checks

  • Full suite: 1801 passed, 13 skipped, 0 failed.
  • osiris --help still works.
  • Ruff E/F/W findings on osiris_repl.py: 30 before, 28 after. No new findings.

🤖 Generated with Claude Code

…surements; add ledger benchmark

osiris_repl.py
- The header listed "11D CRSM", "theta_lock = 51.843" torsion mechanics, CCCE
  "consciousness scoring" and a "Merkle cryptographic proof" as features.
  Rewritten to say what the module wires together; the historical constants
  stay, each marked refuted or unsupported, pointing at the claims register.
- The introspection report printed "CUSUM Drift Anomaly Score = 0.012" and
  "Strategy Diversity Index = 0.942" as literals; they now say "not measured
  in this check". "(Phase-Lock θ=51.843°)" dropped from the coherence line.
- argparse description no longer says "Sovereign Truth Architecture / Apex REPL".

benchmarks/bench_ledger_append.py (+ results_ledger_append.json)
- Append latency, verify time and five tamper cases on a temporary journal.
  2,000 appends: median 0.060 ms, p99 0.164 ms. edit_field, edit_hash,
  delete_middle, swap_adjacent detected; truncate_tail NOT detected -- a hash
  chain alone cannot see a removed tail; that needs an externally anchored
  (e.g. token-signed) tip. Reported as is.

Full suite: 1801 passed, 13 skipped, 0 failed. Ruff E/F/W on osiris_repl.py:
30 -> 28 (no new findings).

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@quantum-osiris
quantum-osiris merged commit eae1198 into main Nov 10, 2026
10 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant